Deploying Your First AI Agent

The Age of Autonomous Agents

Large Language Models (LLMs) are great at answering questions, but Agents can take actions. By giving an LLM access to tools (like a calculator, web search, or a database), you empower it to solve complex, multi-step problems.

In this tutorial, we will build a simple agent using LangChain and deploy it to a Vercel Edge Function.

Prerequisites

• Node.js installed
• An OpenAI API Key
• A Vercel account

Step 1: Project Setup

Create a new Next.js project:

npx create-next-app@latest ai-agent
cd ai-agent
npm install @langchain/openai langchain @langchain/core

Create a .env.local file and add your API key: OPENAI_API_KEY=sk-...

Step 2: Creating the Agent Route

Create a new file at app/api/chat/route.ts. We will use Next.js Route Handlers configured for the Edge runtime.

import { ChatOpenAI } from "@langchain/openai";
import { AgentExecutor, createOpenAIFunctionsAgent } from "langchain/agents";
import { Calculator } from "langchain/tools/calculator";
import { ChatPromptTemplate } from "@langchain/core/prompts";

export const runtime = 'edge';

export async function POST(req: Request) {
  const { messages } = await req.json();
  
  // 1. Initialize the model
  const model = new ChatOpenAI({ temperature: 0, modelName: "gpt-4-turbo" });
  
  // 2. Define tools
  const tools = [new Calculator()];
  
  // 3. Create the prompt
  const prompt = ChatPromptTemplate.fromMessages([
    ["system", "You are a helpful mathematical assistant."],
    ["human", "{input}"],
    ["placeholder", "{agent_scratchpad}"],
  ]);
  
  // 4. Initialize agent
  const agent = await createOpenAIFunctionsAgent({ llm: model, tools, prompt });
  const executor = new AgentExecutor({ agent, tools });
  
  // 5. Execute
  const result = await executor.invoke({ input: messages[messages.length - 1].content });
  
  return Response.json({ output: result.output });
}

Step 3: Deployment

Deploying is as simple as pushing your code to GitHub and importing the repository into Vercel. Because we specified export const runtime = 'edge', Vercel will automatically deploy this route handler to their global edge network, ensuring blazing-fast response times globally.

Remember to add your OPENAI_API_KEY to the Environment Variables section in your Vercel project settings!